model_619006684_beit_xlarge.py

Model Overview

A xlarge-scale implementation of the beit architecture, built for classification tasks.

Architecture

• Architecture: beit
• Scale: xlarge
• Attention: sparse
• Fusion strategy: co attention
• Task head: classification
• Activation: swish
• Normalization: instancenorm
• Initialization: kaiming normal

Training

• Optimizer: novograd
• LR scheduler: step

Files

• model_619006684_beit_xlarge.py — main artifact of this repository

License

See the license field above.